Text-Independent Speaker Identification using Soft Channel Selection in Home Robot Environments

With the aim of achieving the best possible speaker identification rate in a distant-talking environment, we developed a multiple microphone-based text-independent speaker identification system using soft channel selection. The system selects and combines the identification results based on the reliability of an individual channel result using a single perceptron. Thus, it allows for user-customized service with high identification accuracy in home robot environments. From the experimental results, it is shown that the proposed system is effective in a distant-talking environment, thereby providing a speech interface for a wide range of potential hands-free applications in a ubiquitous environment.